I/Q (In-phase/Quadrature) modulators and demodulators are widely used in digital communications systems. I/Q demodulators are abundantly discussed in the technical literature. See, for example, Behzad Razavi, RF Microelectronics, Prentice Hall (1998) and John G. Proakis, Digital Communications, McGraw-Hill (1995). There exists also patent art related to the technology of I/Q modulation and demodulation: U.S. Pat. No. 5,974,306, entitled “Time-Share I/Q Mixer System With Distribution Switch Feeding In-Phase and Quadrature Polarity Inverters” to Hornak, et al.; U.S. Pat. No. 5,469,126, entitled “I/Q Modulator and I/Q Demodulator” to Murtojarvi.
Examples of system applications that incorporate and standardize I/Q modulation and demodulation include the GSM (Global System for Mobile Communications), IS-136 (TDMA), IS-95 (CDMA), and IEEE 802.11 (wireless LAN). I/Q modulation and demodulation have also been proposed for use in the Bluetooth wireless communication systems.
Bluetooth is a low-power radio technology being developed with a view to substituting a radio link for wire and cable that now connect electronic devices, such as personal computers, printers and a wide variety of handheld devices, including palm-top computers, and mobile telephones. The development of Bluetooth began in early 1998 and has been promoted by a number of telecommunications and computer industry leaders. The Bluetooth specification is intended to be open and royalty-free and is available to potential participants as a guide to the development of compatible products.
The Bluetooth system operates in the 2.4GHz ISM (Industrial, Scientific, Medical) band, and devices equipped with Bluetooth technology are expected to be capable of exchanging data at speeds up to 720 Kbs at ranges up to 10 meters. This performance is achieved using a transmission power of 1 mw and the incorporation of frequency hopping to avoid interference. In the event that a Bluetooth-compatible receiving device detects a transmitting device within 10 meters, the receiving device will automatically modify its transmitting power to accommodate the range. The receiving device is also required to operate in a low-power mode as traffic volume becomes low, or ceases altogether.
Bluetooth devices are capable of interlinking to form piconets, each of which may have up to 256 units, with one master and seven slaves active while others idle in a standby mode. Piconets can overlap, and slaves can be shared. In addition, a form of scatternet may be established with overlapping piconets, thereby allowing data to migrate across the networks.
An example of a Bluetooth-compliant digital communications receiver that incorporates an I/Q demodulator is depicted in FIG. 1. As may be seen from FIG. 1, the receiver includes an antenna 10 that intercepts a transmitted RF signal. The signal received by antenna 10 is filtered in a RF bandpass filter (BPF) 11. BPF 11 may be fixed-tuned or tunable and will have a nominal center frequency at the anticipated RF carrier frequency. The bandwidth of BPF will be designed as appropriate to the overall receiver system design requirements and constraints. One salient purpose of BPF 11 is to effect rejection of out-of-band RF signals, that is, rejection of signals at frequencies other than the frequency of the desired RF carrier. Front-end selectivity is an important factor in minimizing the receiver's susceptibility to intermodulation and cross-modulation interference. In addition, and contextually more relevant, BPF 11 selectivity contributes to the image-rejection characteristics of the receiver.
In general, image rejection refers to the ability of the receiver to reject responses resulting from RF signals at a frequency offset from the desired RF carrier frequency by an amount equal to twice the intermediate frequency (IF) of a dual-conversion receiver. For example, if the desired RF signal is at 100 MHz, and the receiver IF is 10.7 MHz, than the receiver local oscillator (LO) will be tuned to 89.3 MHz. However, as is well known to those skilled in the art, the receiver will also exhibit a response to undesired RF signals at frequency 10.7 MHz below the LO frequency, that is 78.6 MHz. The receiver's response to the 78.6 MHz signal is referred to as the image response, because the image signal resides at a frequency opposite the LO frequency as the desired RF carrier, and offset from the LO frequency by the magnitude of the IF.
Referring still to FIG. 1, the output of BPF 11 is coupled to the input of a low-noise amplifier (LNA) 12. LNA 12 is designed to raise the level of the input RF signal sufficiently to effectively drive the receiver's mixer circuitry. In addition, LNA 12 largely determines the receiver's noise figure.
The output of LNA 12 is coupled to the receiver's mixer/demodulator functional block. The mixer/demodulator includes a quadrature demodulator, including I demodulator 13 and Q demodulator 14. As is commonplace in contemporary receiver design, the receiver incorporates a digital, frequency-synthesized LO function, performed by a voltage-controlled oscillator (VCO) 15, driven by a phase-locked loop (PLL) 16. For a comprehensive exposition of digital frequency-synthesis techniques, see William F. Egan, Frequency Synthesis by Phase Lock, John Wiley & Sons, Inc., (2000). The LO signal is coupled to an input of phase-shifter 17. In a manner well understood by artisans, phase-shifter 17 delivers an in-phase version of the LO, LO_I signal 13a, to I demodulator 13 and a quadrature (90° phase shifted) version of the LO, LO_Q signal 14a, to Q demodulator 14. The respective demodulated outputs of demodulators 13 and 14 constitute, respectively, the demodulated I and Q signals.
An ideal I/Q demodulation receiver, as described above, is theoretically capable of infinite image rejection. However, the theoretical assumption is predicated on perfectly matched I and Q channels. Because state-of-the art semiconductor device design and fabrication does not admit of perfect matching between devices, even devices on the same die, some degree of mismatch between the I and Q channels is inevitable. In fact, the mismatch between devices on a semiconductor wafer is known to be dependent on the physical size of the devices. This dependency may be predicted by the following relationships that quantify the standard deviation in threshold voltage σV1, and β, σβ, for a MOS device:
                                                                                      σ          V1                =                              30            ⁢                                                  ⁢                          (                              millivolt                -                micrometer                            )                                                          W              ×              L                                                      and                                                                                                                                                                σ            β                    =                                    0.09              ⁢                                                          ⁢                              (                micrometer                )                                                                    W                ×                L                                                    ,        where            
W×L is total area occupied by the device on the semiconductor die.
As is immediately apparent from the above, deviations in critical CMOS device parameters vary inversely with the area occupied by the device. Because lower frequencies of operation permit larger device geometries, mismatch in a receiver IF section tends to be ameliorated as the IF is reduced.
It has been empirically determined that contemporary semiconductor fabrication processes result in I channel and Q channel matching that limits image rejection to approximately 30 to 35 db. In systems implemented with CMOS technology, virtually mandatory when power consumption is a paramount design consideration, not even this level of performance is realizable. This detriment derives from the fact that CMOS devices tend to demonstrate less favorable matching characteristics. Given that a 35 db image rejection specification is considered marginal for most digital communication receiver applications, the problems confronted in a CMOS-based design are glaringly apparent.
Accordingly, what is desired is a solution that enhances the image-rejection performance of digital communication receivers that are implemented with integrated circuit technology. Although the solution is not limited in applicability of designs implemented in CMOS technology, the invention is particularly advantageous in that context.